Journal of Educational Psychology
2003, Vol. 95, No. 1, 66 –73
Copyright 2003 by the American Psychological Association, Inc.
0022-0663/03/$12.00 DOI: 10.1037/0022-0663.95.1.66
Accuracy of Metacognitive Monitoring Affects Learning of Texts
Keith W. Thiede, Mary C. M. Anderson, and David Therriault
University of Illinois at Chicago
Metacognitive monitoring affects regulation of study, and this affects overall learning. The authors
created differences in monitoring accuracy by instructing participants to generate a list of 5 keywords that
captured the essence of each text. Accuracy was greater for a group that wrote keywords after a delay
(delayed-keyword group) than for a group that wrote keywords immediately after reading (immediatekeyword group) and a group that did not write keywords (no-keyword group). The superior monitoring
accuracy produced more effective regulation of study. Differences in monitoring accuracy and regulation
of study, in turn, produced greater overall test performance (reading comprehension) for the delayedkeyword group versus the other groups. The results are framed in the context of a discrepancy-reduction
model of self-regulated study.
accurately monitored their learning. These findings led them to
conclude that “memory monitoring does not make a valuable
contribution to memory” (p. 212).
Dunlosky and his colleagues (Dunlosky & Connor, 1997; Dunlosky & Hertzog, 1997) have examined the relation between monitoring accuracy and memory by comparing the monitoring accuracy of groups known to differ in memory performance (i.e., older
adults vs. younger adults). Groups that differed in performance did
not differ in monitoring accuracy. Thus, differences in monitoring
accuracy cannot account for the differences in memory performance typically observed between older versus younger adults. In
terms of the relation between monitoring accuracy and memory
performance, these studies may suggest a weak link between these
variables.
Although the aforementioned studies suggest a weak relation
between monitoring accuracy and test performance (see also Kelly,
Scholnick, Travers, & Johnson, 1976), Maki and Berry (1984)
demonstrated that monitoring accuracy was greater for participants
who scored above the median on a test than for those who scored
below the median, which suggests there is a relation between
monitoring accuracy and test performance. Yet, on the basis of a
recent review of the literature, Pressley and Schneider (1997)
concluded that there is no clear evidence of a relation between
prediction (monitoring) accuracy and test performance.
One reason that researchers may have failed to show a relation
between monitoring accuracy and test performance is that they
have not evaluated the causal relations between these variables.
That is, monitoring accuracy may be important because it provides
information to guide self-regulation of study, and regulation affects test performance. Thus, the effect of accurate monitoring can
only be observed if individuals are free to use this information to
regulate their study. When regulation is controlled by the experimenter (as in the investigation by Begg et al., 1992, where study
time was held constant across items), the potential influence of
monitoring on learning will be minimal. By contrast, one might
expect monitoring to affect performance if individuals are allowed
to differentially allocate study time on the basis of their monitoring
of learning.
The investigations that examined differences in monitoring accuracy between groups known to differ in memory ability are
Many models of self-regulated learning can be classified as
discrepancy-reduction models (e.g., Butler & Winne, 1995; Dunlosky & Hertzog, 1997; Dunlosky & Thiede, 1998; Hyland, 1988;
Koriat & Goldsmith, 1996; Le Ny, Denhière, & Le Taillanter,
1972; Nelson & Narens, 1990; Powers, 1973; Thiede & Dunlosky,
1999). According to these models, a person begins study by setting
a desired state of learning for the to-be-learned material. As the
person studies, he or she monitors how well the material has been
learned to determine the current state of learning. If the current
state of learning meets or exceeds the desired state of learning, the
person will terminate study. By contrast, if the current state of
learning has not reached the desired state of learning, the person
will continue to study the material (by either allocating additional
study time to the material or selecting the material for restudy).
During restudy, the person monitors learning and compares the
current state of learning with the desired state of learning. The
person will continue to study until the perceived discrepancy
between the current state of learning and the desired state of
learning reaches zero.
According to these models, accurate metacognitive monitoring
will produce more effective regulation, and this in turn will produce improved learning. However, as noted by Cavanaugh and
Perlmutter (1982), there is no strong empirical evidence linking
monitoring accuracy or self-regulation to measures of learning. For
instance, Begg, Martin, and Needham (1992) examined the relation between monitoring accuracy and test performance. They
found that when items were presented once for study, test performance was actually greater for a group of participants that less
accurately monitored their learning than for a group that more
Keith W. Thiede and Mary C. M. Anderson, Department of Educational
Psychology, University of Illinois at Chicago; David Therriault, Department of Psychology, University of Illinois at Chicago.
This research was supported by James S. McDonnell Foundation Grant
98-7 CSEP-EDU.02. Thanks to John Dunlosky for his comments on a draft
of this article.
Correspondence concerning this article should be addressed to Keith W.
Thiede, Department of Educational Psychology (m/c 147), University of
Illinois at Chicago, 1040 West Harrison Street, Chicago, Illinois 606077133. E-mail: kthiede@uic.edu
66
MONITORING AND SELF-REGULATION
important because they established that differences in test performance might not necessarily be due to differences in accuracy.
However, evaluating whether groups that differed in test performance also differed in accuracy is not the same as evaluating
whether groups that differed in monitoring accuracy differed in
test performance. That is, these investigations did not examine
whether monitoring accuracy affects test performance. To do this,
it is critical to compare test performance between groups that differ
in monitoring accuracy.
A major goal of the present investigation was to evaluate the
role of monitoring accuracy in learning. To do this, we used an
experimental manipulation that produced different levels of monitoring accuracy. We then examined how differences in accuracy
were related to regulation of study (selection of texts for restudy)
and in turn how these differences influenced overall learning.
Factors Influencing the Accuracy of Metacognitive
Monitoring
Monitoring accuracy can be improved in a variety of ways. For
instance, accuracy improves when a person monitors learning after
a delay rather than immediately after studying an item (Dunlosky
& Nelson, 1992; Nelson & Dunlosky, 1991), when items are
actively generated during study rather than passively read during
study (Mazzoni & Nelson, 1993) and when monitoring occurs after
a practice test of the material (King, Zechmeister, & Shaughnessy,
1980; Lovelace, 1984; Shaughnessy & Zechmeister, 1992).
The previous studies involved monitoring learning during an
associative learning task. By contrast, in our investigation, we
examined monitoring learning during reading of texts. This is not
an insignificant methodological difference, given that baseline
levels of monitoring accuracy for texts are quite low (Glenberg,
Sanocki, Epstein, & Morris, 1987; Maki, 1998) and that most
attempts to improve comprehension monitoring have produced
less than impressive results (cf. Rawson, Dunlosky, & Thiede,
2000, who achieved relatively high levels of monitoring accuracy
by instructing participants to reread texts prior to rating comprehension). For example, Maki and Serra (1992) showed that practice tests had only a modest effect on monitoring accuracy—and
only when the practice tests were identical to the eventual tests of
comprehension. Unlike this study that showed practice tests have
only a small effect on relative accuracy, others have shown that
practice tests improve absolute accuracy of monitor (e.g., Ghatala,
Levin, Foorman, & Pressley, 1989; Pressley, Snyder, Levin, Murray, & Ghatala, 1987).
On the basis of recent work by Thiede and Anderson (2003),
who showed that summarizing texts improved monitoring accuracy, we created differences in monitoring accuracy by instructing
participants to generate a list of keywords that captured the essence
of a text prior to rating comprehension. That is, participants read
six texts. After reading the texts, they wrote a list of five keywords
for each text. They then rated their comprehension of each text and
took a comprehension test for each text. In a pilot study, generating
keywords (vs. not generating keywords) improved monitoring
accuracy— however, only when keywords were generated after a
delay (not immediately after reading). Thiede and Anderson used
activation theories of text comprehension (Britton & Gülgöz,
1991; Fletcher, van den Broek, & Arthur, 1996; van den Broek,
Risden, Fletcher, & Thurlow, 1996) to explain why the timing of
67
generation might affect accuracy. According to these theories,
spreading activation occurs during reading; thus, more information
is active in working memory shortly after reading than after a
delay (when activation has decayed). When writing keywords
immediately after reading, a person may have access to a highly
active mental network. Accordingly, the person may have access
to information in short-term memory (STM) to use in writing
keywords even for a text that was not well understood. That is, for
less understood texts, the person may base keywords on extraneous
information activated during reading or on information contained
in the text that is active in STM. However, this information (in
STM) may not be accessible after the mental network has decayed
at the time of the test of comprehension (for a similar discussion of
how monitoring information from STM versus long-term memory
[LTM] may affect monitoring accuracy of associative memory, see
Dunlosky & Nelson, 1992).
The key is that a person may have access to information during
keyword generation even for texts that were not well understood,
so the process of writing keywords of well-understood texts versus
less understood texts may seem quite similar immediately after
reading. Therefore, writing keywords immediately after reading
may produce a set of homogeneous cues for judging comprehension that may not help discriminate well-understood texts from less
understood texts. Moreover, these cues may not be indicative of
test performance, given that the test occurs after a delay; therefore,
one might expect poor monitoring accuracy when keywords are
written immediately after reading. By contrast, activation of the
mental network for a text may have decayed for participants when
writing keywords after a delay, and a person may have access to
only that information retrieved from LTM when writing keywords.
Thus, for a less understood text, the person may have little to draw
on when writing keywords; whereas, for a well-understood text,
the person may retrieve much information during keyword generation. Accordingly, writing keywords after a delay may produce a
set of heterogeneous cues for judging comprehension that may
highlight differences between well-understood texts and less understood texts. Moreover, these cues are likely highly indicative of
test performance because both keyword generation and tests occur
after a delay and are based on retrieval of information from LTM;
therefore, one might expect high levels of monitoring accuracy
when keywords are written after a delay.
In the present experiment, we included three groups. One group
wrote keywords after a delay filled by reading other texts (the
delayed-keyword group), one wrote keywords immediately after
reading a text (the immediate-keyword group), and one did not
write keywords (the no-keyword group). The focus of this investigation was on the relation between monitoring accuracy, regulation of study, and test performance. We will evaluate possible
explanations for the effect of generating keywords on accuracy
elsewhere.
We hypothesized that monitoring accuracy would be greater for
the delayed-keyword group than for the other groups. Moreover,
we hypothesized that the superior accuracy would lead to more
effective regulation of study (i.e., participants would choose to
restudy less learned texts over better learned texts to a greater
degree) than in the other groups, and this would produce greater
test performance for the delayed-keyword group than for the other
groups.
THIEDE, ANDERSON, AND THERRIAULT
68
Method
Subjects, Design, and Materials
Sixty-six students enrolled in a psychology or educational psychology
course at the University of Illinois at Chicago were randomly assigned to
three groups (delayed keyword, immediate keyword, or no keyword) by
order of appearance. They each received $20 for their participation.
The texts were seven expository texts taken from encyclopedias on
different topics (i.e., communication styles of men vs. women, the effects
of alcohol on sleep, experimental design, intelligence and IQ tests, stress,
Norse settlements, and World War II naval warfare). The texts ranged in
length from 1,118 words to 1,595 words, and ranged in Flesch–Kincaid
readability scores from 9.5 to 12.0. A sample text is provided in the
Appendix.
Procedure
An overview of the experimental procedure is presented in Figure 1. All
participants were instructed that they would read texts, rate their comprehension for each text, and then answer test questions for each text. Participants were also instructed that they might be asked to write a list of
keywords that captured the essence of a text. These instructions included an
example of keywords (i.e., for a text on the Titanic, one might write
iceberg, shipwreck, tragedy, etc.), but there was no formal training on how
Figure 1. Overview of the experimental procedure. Read 1 indicates the
time that the first text was presented for reading, Read 2 indicates the time
that the second text was presented for reading, and so on. Keyword 1
indicates the time that keywords were produced for the first text, Keyword 2 indicates the time that keywords were produced for the second text,
and so on.
to generate keywords. All participants, including the no-keyword group,
were given pen and paper. Following the instructions, participants were
asked to make an ease-of-learning judgment for each text. The ease-oflearning judgment was prompted with the title of the text at the top of the
screen and the query “How easily do you think you could learn the
information from a passage on the topic listed above? 1 (very difficult) to 7
(very easy).” After making an ease-of-learning judgment for each text,
participants read the sample text, rated their comprehension of the text, and
answered the sample questions. Participants were encouraged to ask questions about the procedure during the practice trial.
For the critical trials, the order of text presentation was randomized anew
for each participant. Participants in the no-keyword group first read the six
texts. After reading, they rated their comprehension for each text. The
comprehension rating was prompted with the title of the text at the top of
the screen and the query “How well do you think you understood the
passage whose title is listed above? 1 (very poorly) to 7 (very well).” After
rating their comprehension of the last text, they answered six questions for
each text. Three questions assessed text-based knowledge (details available
within a single paragraph of a text), and three inference questions were
designed to assess knowledge of a person’s situation model (for a detailed
description of various representations of texts and how to assess knowledge
of these representations, see Graesser, Millis, & Zwaan, 1997; Kintsch,
1988). Examples of detail and inference questions are provided in the
Appendix.
The texts were rated for comprehension and tested in the same order
as they were presented for reading. Participants in the delayed-keyword
group read each of the six texts. They were then shown the title of a text
and instructed to write five keywords that captured the essence of that
text. Once they finished writing keywords of the text, they pushed the
return key on the computer, which resulted in the presentation of the
next title and keyword instructions. After writing keywords for the last
text, participants rated their comprehension of each text. After rating
their comprehension of the last text, they answered questions for each
text. Participants in the immediate-keyword group read a text. They
were then shown the title of a text and instructed to write keywords for
that text. Once they finished writing keywords for the text, they pushed
the return key, which resulted in the presentation of the next text. They
read and immediately wrote keywords for each text. After writing
keywords for the sixth text, participants rated their comprehension of
each text. Following the last comprehension rating, they answered
questions for each text.
For each group, after answering the last test question, participants were
presented the number of questions they correctly answered over all six
tests. That is, they received feedback regarding overall performance but not
text specific feedback. Providing explicit text-specific feedback might have
reduced the need for an individual to monitor his or her own comprehension. Given the fact that the focus of this investigation was metacognitive
monitoring and its effect on regulation, we chose to provide feedback on
only overall performance.
Following feedback, participants were then shown an array in which
each cell was filled with the title of a text—the cells were numbered from 1
to 6. Participants selected a text for rereading by typing the number of the
corresponding cell. After a text had been selected, it was eliminated from
the list. They could select zero to six texts for rereading. Texts selected for
rereading were randomized anew and presented for rereading. After rereading the final selected text or after selecting no texts for rereading,
participants were again tested on each text. This time participants answered 12 questions for each text (6 detail questions and 6 inference
questions). Six of these questions were the same as those given on the
previous test, and 6 of the questions were new. After responding to the last
test question, participants were informed of their performance across all six
tests (i.e., total test score).
MONITORING AND SELF-REGULATION
Results and Discussion
In the present research, all differences reported as reliable are
significant at p ⬍ .05. When interactions were significant and we
conducted tests of simple effects, we made a Bonferroni adjustment to maintain a family-wise Type I error rate of .05.
Monitoring Accuracy
As in previous studies (e.g., Glenberg et al., 1987; Maki &
Serra, 1992; Weaver, 1990), monitoring accuracy was operationalized as a Goodman–Kruskal gamma correlation between a participant’s comprehension rating and initial (prerereading) test performance across texts. For each participant, we computed a gamma
correlation. The mean of these intraindividual correlations was
then computed across participants within each group. Throughout
this investigation, we computed a Pearson correlation coefficient
whenever we computed a gamma correlation as the measure of
association. We chose to report only gamma because this is arguably a more appropriate measure of association, given that it is not
affected by an individual’s level of test performance or absolute
threshold of confidence (for further discussion, see Nelson, 1984).
However, please note that throughout this investigation these measures of association led to convergent conclusions.
The mean correlation was reliably greater than zero for all of the
groups (ts ⬎ 2.40): Comprehension ratings were predictive of
subsequent test performance. More important, there was a reliable
difference in monitoring accuracy across groups, F(2, 63) ⫽ 4.07,
MSE ⫽ 0.26. As seen in Figure 2, accuracy was substantially
higher for the delayed-keyword group than for the other groups.
69
Moreover, the accuracy of the delayed-keyword group was as high
as any other reported in the literature (only Rawson et al., 2000,
and Weaver & Bryant, 1995, showed monitoring accuracy near the
level of the delayed-keyword group).
Self-Regulation of Study
According to a discrepancy-reduction model of self-regulated
learning, metacognitive monitoring affects learning by influencing
regulation of study. The idea is that a person who can accurately
discriminate better learned material from less learned material will
more effectively regulate his or her study (see Maki, 1995).
Nelson, Dunlosky, Graf, and Narens (1994) and Thiede (1999)
showed that test performance was greater for students who allocated more study to material perceived as less learned than to
material perceived as better learned. Therefore, we operationalized
regulation of study as the correlation between comprehension
ratings and whether a text was selected for restudy (where 1
denoted a text had been selected, and 0 denoted a text had not been
selected). More effective regulation was indexed by a stronger
negative correlation. Given that monitoring accuracy was greater
for the delayed-keyword group than for the other groups, we
predicted more effective regulation for the delayed-keyword group
than for the other groups.
For each participant, we computed a Goodman–Kruskal gamma
correlation between a participant’s comprehension rating and
whether a text had been selected for restudy across texts. The mean
of these intraindividual correlations was then computed across
participants within each group. As seen in Table 1, participants in
the delayed-keyword group were more likely to select less learned
texts over better learned texts than were either of the other groups,
F(2, 58) ⫽ 3.31, MSE ⫽ 0.41. That is, the delayed-keyword group,
to a greater degree than the other groups, compensated for initial
levels of learning by selecting texts that were less learned for
additional study. Given the superior effectiveness of regulated
study, we predicted greater performance on the final test of comprehension for the delayed-keyword group than for the other
groups. As seen in the right-most column of Table 1, differences in
performance cannot be attributed to differences in the number of
texts restudied, which did not differ across groups, F(2,
63) ⫽ 1.14.
Performance on the Second Test of Comprehension
Figure 2. Mean monitoring accuracy presented by group. Monitoring
accuracy was operationalized as a Goodman–Kruskal gamma correlation
between comprehension ratings and test performance computed across
texts for an individual. The data presented are the means of the intraindividual correlations computed across individuals within groups. Error bars
represent standard errors of the mean.
A major goal of this investigation was to evaluate whether
monitoring accuracy and self-regulation of study influence test
performance. We used a 3 ⫻ 2 analysis of variance to compare
performance among the three keyword groups and across two test
trials (one before restudy and one after restudy—this was a
repeated-measures factor). There was a reliable interaction, F(2,
63) ⫽ 3.27, MSE ⫽ 0.02; therefore, we conducted follow-up tests
of simple effects. Test performance on the first test was essentially
equal for the three groups, F(2, 63) ⬍ 1, (see Figure 3), which
suggests that generating keywords had little impact on initial test
performance. However, test performance on the second test, which
followed regulation of study, was reliably greater for the delayedkeyword group than for the other groups, F(2, 63) ⫽ 3.90,
MSE ⫽ 0.03. This difference was due to a substantial increase in
performance across trials for the delayed-keyword group,
THIEDE, ANDERSON, AND THERRIAULT
70
Table 1
Gamma Correlations Between Comprehension Ratings and
Selection of Texts for Restudy (Regulation of Study),
and the Number of Texts Selected for Restudy
Regulation of
study
Texts selected for
restudy
Group
M
SEM
M
SEM
Delayed keyword
Immediate keyword
No keyword
⫺.79
⫺.35
⫺.36
.11
.17
.15
2.59
2.27
2.86
0.20
0.34
0.27
t(21) ⫽ 4.07. By contrast, performance did not change across trials
for the immediate-keyword group, t(21) ⫽ 1.59, and increased
only marginally for the no-keyword group, t(21) ⫽ 2.03.
The superior test performance was consistent across test items
that appeared on the first test (old items) and those that were
presented for the first time. In particular, the proportion correct on
the old items was greater for the delayed-keyword group (M ⫽ .69,
SEM ⫽ .04) than for the immediate-keyword group (M ⫽ .53,
SEM ⫽ .04) or the no-keyword group (M ⫽ .56, SEM ⫽ .05), F(2,
63) ⫽ 4.11, MSE ⫽ 0.04. Likewise, the proportion correct on the
new items was greater for the delayed-keyword group (M ⫽ .66,
SEM ⫽ .04) than for the immediate-keyword group (M ⫽ .51,
SEM ⫽ .05) or the no-keyword group (M ⫽ .55, SEM ⫽ .05), F(2,
63) ⫽ 3.16, MSE ⫽ 0.04.
The importance of monitoring accuracy in self-regulated study
was indexed by comparison of test performance across groups for
texts that were selected for restudy versus those that were not
selected for restudy. By more accurately monitoring comprehension, the delayed-keyword group was better able to identify texts
for which they had performed poorly on tests and texts for which
they had performed well on tests. They used this information to
select texts for restudy. As seen in the two left-most columns of
Table 2, for texts selected for restudy, performance on the first test
was reliably less for the delayed-keyword group than for the other
groups, F(2, 63) ⫽ 3.15, MSE ⫽ 0.07. Furthermore, for texts not
selected for restudy, performance on the first test was reliably
greater for the delayed-keyword group than for the other groups,
F(2, 63) ⫽ 8.75, MSE ⫽ 0.06.
The superior overall performance on the second test for the
delayed-keyword group was produced by dramatically improving
performance on texts selected for restudy (see the third column of
Table 2) as well as maintaining an existing advantage in test
performance for texts not selected for restudy (see the fourth
column of Table 2). Test performance on both selected and nonselected texts after participants restudied texts was greater for the
delayed-keyword group than the other groups (Fs ⬎ 4.80).
Conclusion
Accurate metacognitive monitoring is critical to learning.
Namely, monitoring provides a basis for making decisions about
what to restudy or how long to study material (i.e., regulation of
study). Metacognitive monitoring is related to regulation of study
(e.g., Mazzoni, Cornoldi, & Marchitelli, 1990; Nelson & Leonesio,
1988), and regulation of study is related to test performance
(Thiede, 1999). These studies are important because they provide
evidence of the relations suggested by models of self-regulated
learning.
By manipulating generation of keywords, we were able to
produce groups that varied widely in metacognitive monitoring
accuracy. Monitoring accuracy was reliably greater for the
delayed-keyword group than for the immediate-keyword group or
the no-keyword group. The higher level of accuracy was associated with more effective regulation of study. That is, to a greater
degree than the other groups, the delayed-keyword group compensated for initial levels of learning with additional study time. The
more effective regulation of study was associated with reliable
increases in overall test performance. Thus, this investigation
provided empirical evidence to support discrepancy-reduction
models of self-regulated learning, with metacognitive monitoring
playing an important role in learning.
Table 2
Test Performance for Texts That Were Selected Versus Texts
That Were Not Selected for Restudy
First test
Selected
Figure 3. Mean performance by group for the first test (taken prior to
rereading selected texts) and the second test (taken after rereading selected
texts).
Second test
Not
selected
Selected
Not
selected
Group
M
SEM
M
SEM
M
SEM
M
SEM
Delayed keyword
Immediate keyword
No keyword
.27
.43
.44
.04
.06
.06
.78
.49
.55
.04
.05
.06
.71
.51
.57
.04
.05
.06
.67
.46
.60
.04
.05
.05
MONITORING AND SELF-REGULATION
More accurate monitoring can lead to more effective regulation,
which can lead to higher levels of test performance—as was the
case with the delayed-keyword group. Lower levels of accuracy,
such as those of the immediate-keyword group and the nokeyword group, can lead to less effective regulation of study,
which can lead to fairly unimpressive gains in performance resulting from restudying texts.
Given the low levels of accuracy typically reported in metacomprehension literature (for a review, see Maki, 1998), it seems likely
that left to their own devices people will not accurately monitor
comprehension during reading. If people are not able to distinguish
what they understood versus what they did not understand, it
seems unlikely that they will differentially allocate study time
across material to improve comprehension. Thus, we need to find
ways to improve monitoring accuracy as a means of improving
reading comprehension.
References
Begg, I. M., Martin, L. A., & Needham, D. R. (1992). Memory monitoring:
How useful is self-knowledge about memory? European Journal of
Cognitive Psychology, 4, 195–218.
Britton, B. K., & Gülgöz, S. (1991). Using Kintsch’s computational model
to improve instructional text: Effects of repairing inference calls on
recall and cognitive structures. Journal of Educational Psychology, 83,
329 –345.
Butler, D. L., & Winne, P. H. (1995). Feedback and self-regulated
learning: A theoretical synthesis. Review of Educational Research, 65, 245–281.
Cavanaugh, J. C., & Perlmutter, M. (1982). Metamemory: A critical
examination. Child Development, 53, 11–28.
Dunlosky, J., & Connor, L. T. (1997). Age differences in the allocation of
study time account for age differences in memory performance. Memory
& Cognition, 25, 691–700.
Dunlosky, J., & Hertzog, C. (1997). Older and younger adults use a
functionally identical algorithm to select items for restudy during multitrial learning. Journal of Gerontology: Psychological Science, 52,
178 –186.
Dunlosky, J., & Nelson, T. O. (1992). Importance of the kind of cue for
judgments of learning (JOL) and the delayed-JOL effect. Memory &
Cognition, 20, 374 –380.
Dunlosky, J., & Thiede, K. W. (1998). What makes people study more? An
evaluation of four factors that affect people’s self-paced study. Acta
Psychologica, 98, 37–56.
Fletcher, C. R., van den Broek, P., & Arthur, E. J. (1996). A model of
narrative comprehension and recall. In B. K. Britton & A. C. Graesser
(Eds.), Models of understanding text (pp. 141–164). Mahwah, NJ:
Erlbaum.
Ghatala, E. S., Levin, J. R., Foorman, B. R., & Pressley, M. (1989).
Improving children’s regulation of their reading PREP time. Contemporary Educational Psychology, 14, 49 – 66.
Glenberg, A. M., Sanocki, T., Epstein, W., & Morris, C. (1987). Enhancing
calibration of comprehension. Journal of Experimental Psychology:
General, 116, 119 –136.
Graesser, A. C., Millis, K. K., & Zwaan, R. A. (1997). Discourse comprehension. Annual Review of Psychology, 48, 163–189.
Hyland, M. E. (1988). Motivational control theory: An integrative framework. Journal of Personality and Social Psychology, 55, 642– 651.
Kelly, M., Scholnick, E. K., Travers, S. H., & Johnson, J. W. (1976).
Relations among memory, memory appraisal, and memory strategies.
Child Development, 47, 648 – 659.
King, J. F., Zechmeister, E. B., & Shaughnessy, J. J. (1980). Judgments of
71
knowing: The influence of retrieval practice. American Journal of Psychology, 93, 329 –343.
Kintsch, W. (1988). The use of knowledge in discourse processing: A
construction-integration model. Psychological Review, 95, 163–182.
Koriat, A., & Goldsmith, M. (1996). Monitoring and control processes in
the strategic regulation of memory accuracy. Psychological Review, 103,
490 –517.
Le Ny, J. F., Denhière, G., & Le Taillanter, D. (1972). Regulation of
study-time and interstimulus similarity in self-paced learning conditions.
Acta Psychologica, 36, 280 –289.
Lovelace, E. A. (1984). Metamemory: Monitoring future recall ability
during study. Journal of Experimental Psychology: Learning, Memory,
and Cognition, 10, 756 –766.
Maki, R. H. (1995). Accuracy of metacomprehension judgments for questions of varying importance levels. American Journal of Psychology,
108, 327–344.
Maki, R. H. (1998). Test predictions over text material. In D. J. Hacker, J.
Dunlosky, & A. C. Graesser (Eds.), Metacognition in educational theory
and practice (pp. 117–144). Hillsdale, NJ: Erlbaum.
Maki, R. H., & Berry, S. L. (1984). Metacomprehension of text material.
Journal of Experimental Psychology: Learning, Memory, and Cognition, 10, 663– 679.
Maki, R. H., & Serra, M. (1992). Role of practice tests in the accuracy of
test predictions on text material. Journal of Educational Psychology, 84,
200 –210.
Mazzoni, G., Cornoldi, C., & Marchitelli, G. (1990). Do memorability
ratings affect study-time allocation? Memory & Cognition, 18, 196 –204.
Mazzoni, G., & Nelson, T. O. (1993). Metacognitive monitoring after
different kinds of monitoring. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 21, 1263–1274.
McGovern, T. H. (1993). Norse settlements. In J. E. Cooke (Ed.), Encyclopedia of the North American colonies (Vol. 1, pp. 105–107). New
York: Scribner.
Nelson, T. O. (1984). A comparison of current measures of feeling-ofknowing accuracy. Psychological Bulletin, 95, 109 –133.
Nelson, T. O., & Dunlosky, J. (1991). When people’s judgments of
learning (JOLs) are extremely accurate at predicting subsequent recall:
“The delayed JOL effect.” Psychological Science, 2, 267–270.
Nelson, T. O., Dunlosky, J., Graf, A., & Narens, L. (1994). Utilization of
metacognitive judgments in the allocation of study during multitrial
learning. Psychological Science, 5, 207–213.
Nelson, T. O., & Leonesio, R. J. (1988). Allocation of self-paced study
time and the “labor-in-vain effect.” Journal of Experimental Psychology: Learning, Memory, and Cognition, 14, 676 – 686.
Nelson, T. O., & Narens, L. (1990). Metamemory: A theoretical framework
and new findings. In G. H. Bower (Ed.), The psychology of learning and
motivation (Vol. 26, pp. 125–141). New York: Academic Press.
Powers, W. T. (1973). Behavior: The control of perception. Chicago:
Aldine.
Pressley, M., & Schneider, W. (1997). Introduction to memory development during childhood and adolescence. Mahwah, NJ: Erlbaum.
Pressley, M., Snyder, B. L., Levin, J. R., Murray, H. G., & Ghatala, E. S.
(1987). Perceived readiness for examination performance (PREP) produced by initial reading of text and text containing adjunct questions.
Reading Research Quarterly, 22, 219 –236.
Rawson, K., Dunlosky, J., & Thiede, K. W. (2000). The rereading effect:
Metacomprehension accuracy improves across reading trials. Memory &
Cognition, 28, 1004 –1010.
Shaughnessy, J. J., & Zechmeister, E. B. (1992). Memory monitoring
accuracy as influenced by the distribution of retrieval practice. Bulletin
of the Psychonomic Society, 30, 125–128.
Thiede, K. W. (1999). The importance of accurate monitoring and effective
self-regulation during multitrial learning. Psychonomic Bulletin & Review, 6, 662– 667.
72
THIEDE, ANDERSON, AND THERRIAULT
Thiede, K. W., & Anderson, M. C. M. (2003). Summarizing can
improve metacomprehension accuracy. Contemporary Educational
Psychology, 28.
Thiede, K. W., & Dunlosky, J. (1999). Toward a general model of selfregulated study: An analysis of selection of items for study and selfpaced study time. Journal of Experimental Psychology: Learning, Memory, and Cognition, 25, 1024 –1037.
van den Broek, P., Risden, K., Fletcher, C. R., & Thurlow, R. (1996). A
“landscape” view of reading: Fluctuating patterns of activation and the
construction of a stable memory representation. In B. K. Britton & A. C.
Graesser (Eds.), Models of understanding text (pp. 165–187). Mahwah,
NJ: Erlbaum.
Weaver, C. A., III. (1990). Constraining factors in calibration of comprehension. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 16, 214 –222.
Weaver, C. A., III, & Bryant, D. S. (1995). Monitoring of comprehension:
The role of text difficulty in metamemory for narrative and expository
text. Memory & Cognition, 23, 12–22.
Appendix
Sample Text and Test Questions
The Viking Age Scandinavians had a lasting impact upon the peoples of
Western Europe. Their settlements, commercial ventures, and raids affected cultures from the Russian plains to the Irish Sea and from northernmost arctic Norway to the Mediterranean. During the Viking period (ca.
790 –1100), Scandinavians also ventured across the North Atlantic, settling
the Shetland and Faroe islands, Iceland, and Greenland and making a brief
appearance on the shores of America. This North Atlantic arm of the
Viking Age expansion connected the eastern and western hemispheres,
and, for a few years at the end of the tenth century, a single language and
culture reached from Kiev to the gulf of St. Lawrence.
By the beginning of the Viking Age, most of Scandinavia was organized
into a maze of local chieftainships. Chieftains were expected to be effective
in protecting their clients and aggressive in pressing for every advantage
for themselves and their supporters in their struggles with rival chieftains.
Traditional law codes (which became increasingly formalized during the
Viking period and were written down soon after) and the independence of
farmer-clients served somewhat as a restraint on chiefly ambition, but
warfare and blood feuds were still commonplace. While Norway, Sweden,
and Denmark were known as geographical terms, nothing resembling a
nation-state (even by eighth-century standards) existed in pre-Viking
Scandinavia.
As wealth from abroad entered Scandinavia, and as Scandinavian merchants, travelers, and mercenaries learned more of the kingdoms of the
outside world, the combination of new resources and new ideas seems to
have sparked increased competition among local chieftains and petty kings.
Agriculture also prospered as a period of warm climate (now known as the
Little Climatic Optimum) lengthened growing seasons in northwestern
Europe. Population seems to have enlarged, which led to the settlement of
the uplands and the extension of Norse farms into arctic Norway. The
expansion of territorial boundaries during the Viking Age provided an
outlet for this growing rural population and yielded new territory for the
losers in the intensifying struggles among chieftains for dominance.
Neither a growing population nor competing chieftains would have
produced the Viking expansion had the means for overseas travel, trade,
and conquest been lacking. Through the efforts of maritime archaeologists,
we know a good deal about Viking period ships and their construction. By
the late eighth century, Scandinavian clinker-built ships had reached a high
level of perfection, combining lightness and shallow draft with great
strength and sea-keeping ability. Viking ships could land on any beach,
penetrate far up rivers, and survive North Atlantic storms on the open sea.
While strong and elegant, the clinker-built Viking ships had two significant limitations. They required a long run of high-quality timber (preferably oak) for the keel and naturally curved timbers for the stem and stern
pieces. Since this quality timber was absent in the North Atlantic islands,
settlers in Iceland and Greenland found it hard to replace oceangoing ships
lost at sea. The Viking design also sharply limited cargo capacity— even
the knarrs (trading vessels) could carry only a fraction of the cargo of the
later carvel-built Hanseatic cogs that came to dominate European commerce in the later Middle Ages. Viking ships could reach distant points, but
they could not carry enough passengers and supplies to ensure a viable
transatlantic foothold. Population movement across the North Atlantic thus
required a chain of settlements, each providing population and resources
for successive ventures westward.
Scandinavian North Atlantic settlement was a gradual process taking
two hundred years to complete. Norse colonists settled the Shetlands and
Orkneys around the year 800 and (according to tradition) Iceland around
AD 874. Greenland was settled from Iceland by Eirik the Red around 985.
Vinland was explored from Greenland and a settlement was attempted by
the sons and daughter of Eirik around the year 1000.
Island chieftains who (like Eirik) had failed in local power struggles
provided the ships and capital to sponsor further voyages of exploration
and settlement. Unsuccessful farmers and dissatisfied younger siblings
from successively filled island ecosystems provided the bulk of the personnel. The first settlers in a new land had the ritually important right to
name the landscape and economically vital right to claim the best pasture
and hunting grounds. As prime grazing is often patchy and limited in the
North Atlantic islands, this initial division of resources set the stage for
increasing economic and social hierarchy in later generations.
During the eleventh and twelfth centuries, the Scandinavian North
Atlantic enjoyed modest prosperity. Island populations seem to have stabilized at low levels; Iceland’s population was probably between thirty
thousand and sixty thousand, and Greenland’s was six thousand at most.
While state formation was taking place in the Scandinavian homelands, the
more distant North Atlantic islands seem to have maintained a somewhat
archaic chiefly oligarchy. Christianity had spread as far as Greenland by
the year 1000, and most Scandinavians were at least nominally Christian by
1100. Chiefly competition was now conducted through the endowment of
churches and monastic houses as well as by the traditional sheep stealing
and house burning. In Iceland and probably Greenland, sagas and family
histories were being composed, and poets and skalds from the North
Atlantic were still in demand in continental courts.
Along with prosperity came the beginnings of decline. Iceland’s chiefly
dominance struggles had thrown up six great families whose escalating
warfare increasingly exhausted local resources. Overgrazing in many areas
triggered massive and irreversible soil erosion, turning whole districts into
rocky wasteland. After 1250, volcanic eruptions coupled with the end of
the favorable weather of the Little Climatic Optimum added to man-made
disaster, and increasing numbers of North Atlantic farmers slipped from
freeholder to tenant status.
After 1264, Iceland and Greenland became part of the Norwegian
kingdom just as that kingdom was about to enter a long period of decline.
Their local oceangoing ships long lost, the settlers of the western Atlantic
depended upon continental merchants to carry their trade. Icelanders bitterly complained that the promised six ships per year seldom arrived, and
MONITORING AND SELF-REGULATION
it seems to have taken a papal letter five years to reach Greenland. The
eastern Atlantic settlements in the Shetlands and northern Scotland were
luckier, as they were becoming increasingly integrated into the stock fish
trade through the Hanseatic League.
The late thirteenth and the fourteenth centuries saw accelerated decline
in the western North Atlantic. The onset of the Little Ice Age (ca. 1250 –
1860 in the North Atlantic) crippled farming, and economic hardship in
Norway affected transatlantic trade. Literature declined, and the populations of Iceland and Greenland became locked in a struggle for bare
survival. By the later Middle Ages, the Norse North Atlantic was no longer
the cutting edge of an expanding European population but a demoralized
and isolated backwater.
Detail question: Which was settled first by Norse colonists?
A.
Greenland
B.
Iceland
C.
The Shetlands*
D.
Vinland
73
Inference question: What modern group is most like the chieftainships of
the Viking period?
A.
Urban gangs*
B.
Midwest farmers
C.
State legislators
D.
College students
Note. Asterisks indicate correct answer. Text (excluding questions) from
“Norse Settlements,” by T. H. McGovern, in Encyclopedia of the North
American Colonies (Vol. 1, pp. 105–107), edited by J. E. Cooke, 1993,
New York: Charles Scribner’s Sons. Copyright 1993 by Charles Scribner’s
Sons. Reprinted by permission of The Gale Group.
Received July 24, 2001
Revision received March 8, 2002
Accepted March 14, 2002 䡲